4b1afb1073
Add Obsidian vault to the swarm-common virtiofs share for access from zap VM and other VMs. Contains agent memory, notes, and infrastructure documentation. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
1.8 KiB
1.8 KiB
title, area, tags, created, updated, status, related
| title | area | tags | created | updated | status | related | ||||
|---|---|---|---|---|---|---|---|---|---|---|
| Docker Services | infrastructure |
|
2026-03-18 | 2026-03-18 | active |
|
Docker Services
All services run on the LAN host (192.168.153.113). See Infrastructure/Architecture for networking context.
Services
SearXNG
- Port:
18803 - Role: Local-first web search (preferred over Brave for privacy)
- API: JSON enabled
- Smoke test:
skills/searxng-local-search/scripts/smoke.sh openclaw
LiteLLM Proxy
- Port:
18804 - Role: OpenAI-compatible gateway routing to all model providers (Copilot, ZAI, local)
- Sync: Auto-synced every 12h via Infrastructure/Automation/Cron Jobs
Whisper Server
- Port:
18801 - Role: Local audio transcription (whisper.cpp)
- Skill:
whisper-local-safe
Brave MCP
- Port:
18802 - Role: Brave Search MCP server — fallback / second opinion
- Note: Free plan rate-limited; serialize requests, avoid parallel bursts
n8n-agent
- Port:
18808 - Role: Automation workflows for tasks with native n8n nodes
- Workflows: See Infrastructure/Automation/n8n Workflows
Ollama
- Port:
18807 - Role: Local embeddings server
- Model:
nomic-embed-text(768 dims, loaded forever) - Used by: OpenClaw memory search
llama.cpp (Qwen2.5-14B)
- Port:
18806 - Model:
Qwen2.5-14B-Instruct-Q4_K_M.gguf - Role: Private/offline LLM, tool-calling capable (verified with
--jinja) - Context: 32768 tokens, max output 8192
- LiteLLM alias:
litellm/qwen2.5-14b-local
MinIO
- Endpoint:
192.168.153.253:9000 - Bucket:
zap - Role: Object storage for OpenClaw backups
- Versioning: Enabled, 90-day noncurrent retention